**** BEGIN LOGGING AT Wed Jan 03 02:59:58 2007 Jan 03 03:45:01 hi Jan 03 03:45:01 hi Jan 03 03:45:01 hi Jan 03 03:56:59 diet-x11 fails in an Angstrom build too Jan 03 03:58:00 interestingly, though, bitbake -k works as expected when doing bitbake -k world, but doesn't when doing bitbake -k slugos-packages. So it might be a slugos-specific problem instead of a bitbake -k problem. Jan 03 06:00:07 hrw|gone: how can I turn off eabi? Jan 03 06:02:14 JustinP: Change the target from linux-gnueabi, I'd imagine.. hrw|gone's probably better equipped to give a concrete answer Jan 03 06:26:03 morning Jan 03 06:26:10 JustinP: use angstrom-2007.1-oabi distro Jan 03 06:28:13 monsti: read kbdd page in handhelds.org wiki Jan 03 06:33:36 hrw: is angstrom softfloat? Jan 03 06:34:28 yes Jan 03 06:39:32 ok, thanks Jan 03 06:42:54 hrw: thx Jan 03 06:43:09 np guys Jan 03 06:43:57 hrw: it took me 3 days to snoop that braindead protocol :/ Jan 03 06:44:35 hrw: I'm attempting to help debug a problem with gpsd and garmin binary protocol. Works in angstrom but not in OZ. Turning off softfloat didn't help in OZ. Seems to be a compiler issue. Jan 03 06:45:29 monsti: there is also zkbdd which was hacked to use lua for definitions of keyboards but this fork is nearly not maintained Jan 03 06:45:44 monsti: Ir keyboard? do you use 2.6 kernel? Jan 03 06:45:52 no :) Jan 03 06:46:01 i wannt to use kbdd for psp := Jan 03 06:46:52 and of couse i just got some braindead ir keyboard - so twice the fun :) Jan 03 06:48:08 i think i just replace dev_uinput.c for a playstation version Jan 03 06:51:54 monsti: maybe rather extend it and send patch upstream? Jan 03 07:05:21 i am writiing a mail to one of the guys in the wiki Jan 03 07:07:01 bluelightning is proper iirc Jan 03 07:07:20 you rang? :) Jan 03 07:09:05 ;D Jan 03 07:11:32 damn - i really hate hardware with no specs :=) Jan 03 07:44:40 quit Jan 03 07:45:04 history |grep links Jan 03 07:45:09 history |grep lynx Jan 03 07:57:00 cu Jan 03 08:00:25 03hrw 07org.oe.dev * r122bf822... 10/ (1 conf/machine/palmtx.conf): palmtx: added machine description for Palm Tungsten X - close #1681 Jan 03 08:00:31 03hrw 07org.oe.dev * ra17d766a... 10/ (1 conf/machine/palmld.conf): palmld: fix kernel name Jan 03 08:00:37 03hrw 07org.oe.dev * rc1c78bb1... 10/ (11 files in 5 dirs): linux-hackndev 2.6: added kernel recipe for Palm Tungsten X and LiveDrive - partially close #1683 Jan 03 08:25:40 morning all Jan 03 08:27:58 morning Jan 03 08:31:23 morning Jan 03 08:31:43 morning Jan 03 08:48:40 is there something funky about monotone.openembedded.org today? Jan 03 08:49:30 I pushed/pulled to ewi Jan 03 08:49:44 good morning all Jan 03 08:53:28 morning (office again ;) Jan 03 09:04:37 ~lart unison Jan 03 09:04:37 * ibot pulls out his louisville slugger and uses unison's head to break the homerun record Jan 03 09:05:51 morning, koen: what time zone is mickey on? Jan 03 09:06:13 the german one Jan 03 09:11:48 odd, he seems to be here mostly at night.. Jan 03 09:12:10 koen: monotone.openembedded.org works again? Jan 03 09:13:27 hrw|work: it should Jan 03 09:13:36 hrw|work: it shouldn't have stopped working Jan 03 09:13:52 hrw|work: but it is a bit clogged with a zillion people pulling 170 bogus csets Jan 03 09:14:05 ah.. that ones Jan 03 09:14:16 koen: good that ewi still serves Jan 03 09:14:43 and the one at wolfson Jan 03 09:16:17 I will try to find time to write summary of my angstrom-mega-build Jan 03 09:18:28 koen: yeah, that one still seems to be running Jan 03 09:23:13 03koen 07org.oe.dev * r7b4b2e61... 10/ (3 files in 3 dirs): Jan 03 09:23:13 testdisk_6.5+6.6-WIP.bb: add testdisk. From the homepage: Jan 03 09:23:13 TestDisk is a powerful free data recovery software! It was primarily designed to Jan 03 09:23:13 help recover lost partitions and/or make non-booting disks bootable again when Jan 03 09:23:13 these symptoms are caused by faulty software, certain types of viruses or human Jan 03 09:23:14 error (such as accidentally deleting your Partition Table). Partition table Jan 03 09:23:18 recovery using TestDisk is really easy. Jan 03 09:25:55 If I only had testdisk and photorec a few years ago Jan 03 09:26:27 few years ago I used AmigaOS which had such apps Jan 03 09:26:31 I have used testdisk twice to recover when some stupid installer fscked my partition table (win98, gentoo) Jan 03 09:29:40 4 hours though to analise my disk :-( Jan 03 09:30:18 koen: sorry about the merge csets. Jan 03 09:30:44 rwhitby: why doesn't the autobuilder push the merges? Jan 03 09:30:52 morning all Jan 03 09:30:53 hi Liam Jan 03 09:30:59 yo lrg Jan 03 09:31:06 rwhitby: or even easier, why doesn't it just pick the first head and build from there? Jan 03 09:31:08 hey lrg Jan 03 09:31:18 koen: I could make it push the merge easily - would that be preferable? Jan 03 09:31:32 (I figured it could get into a merge war with another autobuilder ...) Jan 03 09:31:46 building from first head is a good idea though. Jan 03 09:31:46 just picking the first head and build that would be preferable Jan 03 09:31:55 I'll do that. Jan 03 09:32:12 thanks Jan 03 09:32:55 mtn update -r`mtn automate heads |head -n1` Jan 03 09:33:02 I use it in my fetch script Jan 03 09:33:52 I have something similar Jan 03 09:34:05 I wonder if that would be appropriate for the nslu2-linux master makefile in general. then the normal user would never see multiple heads Jan 03 09:35:00 it the easiest solution Jan 03 09:35:30 morning all Jan 03 09:35:42 hrw|work: did the h&d people hack kconfig to parse html? Jan 03 09:35:44 hey RP Jan 03 09:36:51 koen: I'm going to automerge the org.nslu2-linux branches, and do the first head trick on org.oe.dev (leaving it up to OE people to do any multiple head merges) Jan 03 09:36:55 koen: palmld/defconfig? I'll fix it Jan 03 09:37:10 koen: thats my fault Jan 03 09:37:36 hrw|work: and http://www.openembedded.org/repo/org.openembedded.dev/packages/linux/linux-hackndev-2.6/palmtx/init is bogus as well Jan 03 09:37:45 ROOT_DEV=$FILES_MNT/opie-image-v0.8.4-palmtx.rootfs.ext2!?!?! Jan 03 09:39:01 file /bin/busybox ../busybox-1.1.0/busybox 4755 0 0 Jan 03 09:39:03 hmmm Jan 03 09:39:15 I suspect we'll need a initramfs.bbclass Jan 03 09:39:22 :D Jan 03 09:40:32 ~hail pH5 Jan 03 09:40:35 * ibot bows down to pH5 and chants, "I'M NOT WORTHY!!" Jan 03 09:42:09 ibot: yes you are Jan 03 09:42:09 koen: defconfig fixed Jan 03 09:42:09 03hrw 07org.oe.dev * r1b97b656... 10/ (1 packages/linux/linux-hackndev-2.6/palmld/defconfig): linux-hackndev: FIX palmld defconfig Jan 03 09:42:09 hey koen, hrw|work Jan 03 09:42:09 hi ph5 Jan 03 09:48:31 koen: master makefile updated - that should be the end of that problem. Jan 03 09:49:18 koen: what's the general practice for dealing with bitbake world failures on Angstrom/ixp4xxle ? I want to set up the nslu2-linux autobuilder to build that. Jan 03 09:50:31 rwhitby: Do you have any scripts which handle sharing the output from your autobuilder? Jan 03 09:50:55 ahah, why use unison when rsync rulez :-) Jan 03 09:51:29 RP: we used to participate in tinderbox, but it got too unstable and clogged our autobuilder too often. if it is stable know we could try that again. Jan 03 09:51:40 the output packages go to ipkg.nslu2-linux.org/feeds/... Jan 03 09:52:11 mtn: 3 heads on branch 'org.openembedded.dev' Jan 03 09:52:12 nice Jan 03 09:52:58 three heads are a sign of lots of activity - that's a good sign :-) Jan 03 09:53:21 pushing merges Jan 03 09:54:01 (note that nslu2-linux people won't be automatically resolving multiple heads now, due to the change in the master makefile requested above. Jan 03 09:54:34 that used to be automatically done by the master makefile, but now it just chooses the first head. Jan 03 09:55:24 rwhitby: I don't mean the build results, I mean the actual generated images/packages Jan 03 09:56:13 RP: yep, there are all at ipkg.nslu2-linux.org/feeds/ for slugos and unslung. when we get Angstrom to build world successfully for ixp4xxle, the packages will be there too. That site is mirrored at OSUOSL. Jan 03 09:56:43 (not images though, because they contain the intel microcode, so must be behind a click-through like those at www.slug-firmware.net) Jan 03 09:57:57 rwhitby: ok, I just wondered if you had a script for dealing with the image output from OE. I guess I'll have to write one ;-) Jan 03 09:58:46 we just blow away tmp/deploy/images/*.img before building. Jan 03 09:58:59 (to save the autobuilder disk filling up with dated image files) Jan 03 10:00:00 rwhitby: how much space autobuild takes? Jan 03 10:00:34 over 200GB for nslu2-linux, but that includes Optware autobuilds for 10 targets. Jan 03 10:00:54 rwhitby: Its an autobuilder filling up disks thats my problem :). I want to archive the last say 3 sets of images though and share them somewhere outside the build dir... Jan 03 10:01:40 hrw|work: the feeds themselves are only about 35GB. Jan 03 10:02:12 rwhitby: my 'all machines for angstrom' build of bootstrap-images took over 60GB Jan 03 10:03:06 hrw|work: good thing our autobuilder machine has 75GB free for angstrom/ixp4xxle then :-) Jan 03 10:04:21 rwhitby: I usually file bugs when things go boom with builds :) Jan 03 10:04:40 koen: just wondering about things like opie and such for ixp4xxle ... Jan 03 10:04:50 angstrom doesn't support opie Jan 03 10:05:01 ~fish opie Jan 03 10:05:03 * ibot slaps opie around with a large trout Jan 03 10:05:13 hmm - bitbake -k world for angstrom/ixp4xxle builds opie stuff Jan 03 10:05:39 * koen doesn't have any opie stuff in his local tree :D Jan 03 10:05:46 bitbake world is one heck of an undertaking :-/ Jan 03 10:06:17 I stopped building world years ago Jan 03 10:06:21 RP: ah, but I've been chastised before for creating slugos-packages, so what's my alternative ? Jan 03 10:06:34 create a task Jan 03 10:07:04 what's the existing task for all packages that build for Angstrom? Jan 03 10:07:14 there is none Jan 03 10:07:29 so how do you populate your angstrom package feed for your users? Jan 03 10:07:29 rwhitby: You could look at the meta-gpe and meta-opie "packages" as examples of "build xyz packages" Jan 03 10:07:37 'all packages that build ' isn't very usefull Jan 03 10:07:49 rwhitby: build package, test package, upload to feed Jan 03 10:07:59 bah - testing is for users. Jan 03 10:08:01 bitbake world ; rsync is a QA nightmare Jan 03 10:08:49 ok, I'm down with creating a task then. how about task-nas ? Jan 03 10:09:00 (i.e. what taxonomy do you use for task names?) Jan 03 10:10:02 I usually ask myself "what does a user expect from ipkg install task-foo?" Jan 03 10:10:04 rwhitby: You could also mark things which you didn't want to build using something like EXCLUDE_FROM_WORLD_pn-opie-something = "1" and create and exclusions file somewhere Jan 03 10:10:23 RP: MACHINE via env seems to work here Jan 03 10:10:32 koen: Great :) Jan 03 10:10:32 RP: haven't built from scratch, though Jan 03 10:10:52 koen: It should be fine - it certainly works in poky Jan 03 10:10:55 RP: exclusions is a good idea. Jan 03 10:11:05 (does it take regexps?) Jan 03 10:11:17 rwhitby: sadly not Jan 03 10:11:56 RP: got an example of an excludes file? find -name '*exclude*' was unsuccessful. Jan 03 10:12:15 rwhitby: We don't have one. I've hinted about how to do it though ;-) Jan 03 10:12:29 s/about/above/ Jan 03 10:12:38 np - was just looking for stylistic convention to take note of. Jan 03 10:12:53 would it go in conf/distro or packages/meta? Jan 03 10:13:10 (I assume it's called angstrom-excludes.inc?) Jan 03 10:13:32 Its a conf file since it adds things into each packages metadata using the pn- override Jan 03 10:14:22 so conf/distro/angstrom-excludes.conf, included by angstrom-2007.1{-oabi,}.conf ? Jan 03 10:14:44 rwhitby: For now I'd call it something specific to what you're doing - angstrom-ixp4xxle-world-excludes.inc and just add the inc to your local.conf Jan 03 10:15:02 I don't know if koen will want such a thing enabled globally Jan 03 10:15:17 Are there likely to be a whole lot of ixp4xxle specific failures? Jan 03 10:15:43 I don't know... Jan 03 10:16:01 ok, so conf/distro/include/angstrom-ixp4xxle-world-excludes.inc it is Jan 03 10:16:25 rwhitby: If more people want to use it, it can be renamed and generalised Jan 03 10:16:39 nod Jan 03 10:16:53 perhaps have _ixp4xxle as an override in it? Jan 03 10:17:13 rwhitby: If its an ixp4xxle specific problem, yes Jan 03 10:17:16 (then people can just take out the override suffix for general angstrom-excluded packages) Jan 03 10:17:44 RP: I'm thinking of calling it angstrom-world-excludes.inc, and then putting ixp4xxle suffix on all Jan 03 10:17:57 let's not go that route Jan 03 10:17:58 (which can then be removed, say, for all opie stuff) Jan 03 10:18:15 koen: what's your preference? Jan 03 10:18:35 not building world? Jan 03 10:18:44 or using MACHINE_FEATURES = screen Jan 03 10:19:41 and if you're going to build angstrom, packages, why not upload them to the angstrom feed: http://www.angstrom-distribution.org/repo/ ? Jan 03 10:19:57 so you're suggesting I edit all the opie .bb files to only build for machines with a screen? Jan 03 10:20:04 http://www.hrw.one.pl/2007/01/03/2006-timeline/ Jan 03 10:20:15 koen: I'd be happy to upload them anywhere I have write access. Jan 03 10:20:21 rwhitby: libqte would be enough Jan 03 10:20:24 koen: The problem is if you don't have a known good list and you don't have an exclusion list, how do you know what to build in the autobuilder? Jan 03 10:20:43 rwhitby: if libqte will be skipped for !screen images then whole opie and qpe apps will be skipped Jan 03 10:20:47 This is a problem which will come up again :-/ Jan 03 10:21:20 RP: I think koen is saying exclude whole heaps of things using FEATURES, and only use the excludes file as a last resort of other misc failures. Jan 03 10:21:58 * rwhitby hopes he is not putting words in koen's mouth. Jan 03 10:22:30 RP: an exclusion last has to be thought through, especially considering angstrom is building for a lot of machines Jan 03 10:22:31 libqte-x11 would be a valid target even on a slug Jan 03 10:22:47 my ixp4xx machine has a vga plug :) Jan 03 10:22:52 * koen hugs usb-vga Jan 03 10:22:54 I think it will get ugly. Think about python's tk build time dependency for example Jan 03 10:23:37 RP: that one is an example of something which should not be exluded - the diet-x11 package should be fixed instead. Jan 03 10:23:50 (as X libs can be useful on any headless box) Jan 03 10:25:03 koen: your usb-vga is why I want to build as many packages as I can. I can't predict what people will want to do, and don't have the time to respond to every little request one at a time to enable a new package on the autobuilder. Jan 03 10:25:22 (for ixp4xxle) Jan 03 10:25:34 rwhitby: With usb-vga, you could run opie :-/ Jan 03 10:25:57 right. that's why my autobuilder should build opie (assuming it builds for ixp4xxle) Jan 03 10:26:13 RP: iirc the sis doesn't expose a good framebuffer Jan 03 10:26:55 * rwhitby starts angstrom-ixp4xxle-world-excludes.inc while waiting for a good consensus ... Jan 03 10:26:58 rwhitby: you're making one mistake: ixp4xxle isn't the only armv5te machine angstrom supports Jan 03 10:27:17 Ok, so the exclusions list is looking to be as small as possible. I suggest a file specifically for your autobuilder for now which might grow into something more general purpose Jan 03 10:27:43 RP: angstrom-slugos-world-excludes.inc ;-) Jan 03 10:27:54 It'll also be useful for tracking failures as we'll have a list of broken packages in theory... :) Jan 03 10:28:07 rwhitby: right :) Jan 03 10:28:20 * koen would like tinderbox to file a bug when encountering a failure Jan 03 10:28:45 koen: Without some decent bugzilla admins, that would be a nightmare Jan 03 10:29:22 RP: or some other database Jan 03 10:29:39 koen: Some other database would be much better Jan 03 10:29:41 the reports have to be store somewhere Jan 03 10:29:58 * koen wants tarball releases + coverage matrix Jan 03 10:32:05 before the next term starts :/ Jan 03 10:39:52 * koen hugs multimachine Jan 03 10:43:17 RP: is the syntax actually "EXCLUDE_FROM_WORLD_pn-" ? Jan 03 10:47:13 * rwhitby finds a BUILD_OPTIMIZATION_pn-perl = "-O1" example Jan 03 10:48:32 rwhitby: yes Jan 03 10:56:27 RP: KERNEL_RELEASE = "2.6.19-rc4" in 2.6.19+git file?? Jan 03 10:56:40 XorA: ah, yes :-/ Jan 03 10:56:58 RP: not a problem, I was just nicking some stuff from that file :-=) Jan 03 10:57:32 XorA: 2.6.19-rc3 ;-) Jan 03 10:58:01 sorry, 2.6.10-rc2. Jan 03 10:58:07 no... Jan 03 10:58:10 * RP gives up Jan 03 10:58:13 RP: go fetch coffee Jan 03 10:58:35 RP: how about a check like I did for the hh kernels? Jan 03 10:58:54 make OE die if KERNEL_RELEASE and the Makefile don't match Jan 03 10:59:15 koen: If you can make it do that, you can set KERNEL_RELEASE automatically ;-) Jan 03 11:00:18 can you set bitbake vars inside methods? Jan 03 11:00:53 bb.data.setVar() iirc Jan 03 11:01:13 right :) Jan 03 11:01:36 if [ `grep EXTRAVERSION Makefile | grep hh | awk '{print $3}' | sed s/-hh//` != ${HHV} ]; then Jan 03 11:01:37 die "-hh version mismatch" Jan 03 11:01:37 fi Jan 03 11:01:54 You need a python function, not shell Jan 03 11:10:53 Actually, I could even fix reenoo's pet hate with bitbake trunk too... Jan 03 11:13:09 or you could do a release cycle and merge mickeyl's gui stuff Jan 03 11:13:38 koen: I mean by fixing KERNEL_RELEASE once and forall, the problem with KERNEL_VERSION would go away too Jan 03 11:14:31 koen: mickeyl's gui stuff can't be merged until we sort out an ipc. Once mickey digs out some code for me, I'll look at it Jan 03 11:15:49 hmmm Jan 03 11:15:50 http://download.gnome.org/sources/gnome-vfs-obexftp/0.1/ Jan 03 11:15:54 tempting... Jan 03 11:16:31 koen: need gnome-vfs-sdp as well Jan 03 11:19:05 Why aren't KERNEL_VERSION and KERNEL_RELEASE the same thing? Jan 03 11:20:59 gah Jan 03 11:21:11 dnus now behaves on my hx4700, but not on the hx2200 Jan 03 11:21:19 using the same packages Jan 03 11:21:53 * koen blames is on xscale cache problems Jan 03 11:22:23 koen: You have an hx2xxx? Jan 03 11:22:38 ehm h2200 Jan 03 11:22:46 * RP wondered for a minute :) Jan 03 11:24:20 I'm going to axe the KERNEL_RELEASE variable Jan 03 11:38:37 re Jan 03 11:45:56 compulab-pxa270 - who maintain it? Jan 03 11:49:33 dht-walnut.. someone built it for angstrom? Jan 03 11:50:47 koen: efika/angstrom is buildable? Jan 03 11:57:40 hrw|work: it should be Jan 03 11:57:54 hrw|work: I built a angstrom-bootstrap-image for efika yesterday Jan 03 11:58:50 hrw|work: cbrake maintains compulab Jan 03 11:59:15 koen, does it boot? Jan 03 11:59:48 I only chrooted it Jan 03 12:01:23 koen: then probably my buildtree is fscked Jan 03 12:02:26 ~. Jan 03 12:02:42 . Jan 03 12:10:14 .. Jan 03 12:20:37 damn ssh, why does ~. do whats its supposed to sometimes and not others Jan 03 12:25:46 hrw|work: there is a NAND binary driver in the compulab build that is not EABI compatible. Hopefully, we can move to OSS JFFS2 drivers at some point ... Jan 03 12:27:18 cbrake: thats what I suspected. Jan 03 12:28:41 hmm, compulab has an updated kernel patch -- sure wish they would version the files on their site. Jan 03 12:30:07 cbrake: add md5sum to file in SRC_URI Jan 03 12:32:58 ipkg-make-index must die Jan 03 12:34:25 hrw|work: ok, will do Jan 03 12:34:40 "675,858,558 bytes in 22074 files" results in 22M Packages file Jan 03 12:40:31 * XorA chuckles, slightly bigger than collie can store Jan 03 12:41:36 ~kill ipkg-make-index Jan 03 12:41:49 * ibot shoots a ionized fluxneutrino gun at ipkg-make-index Jan 03 12:47:01 It looks like ipkg crashes the 2.6.19 kernel the same as apt? I read were there is a bug that everyone has been tracking down in the 2.6.19 kernel regarding this.. does anyone know if there is a patch yet? Jan 03 12:48:12 * RP mutters about Intel technical support Jan 03 12:48:50 Its the first time I've ever had to deal with a failed processor :-/ Jan 03 12:53:12 lrg: ping Jan 03 12:56:52 koen: interesting mail sent to OEML Jan 03 13:01:05 greentux: pong Jan 03 13:01:11 RP: pxa ? Jan 03 13:01:31 lrg: do You need a zaurus again? I can bring one with me to fosdem Jan 03 13:01:41 greentux, Jan 03 13:01:54 greentux: not atm as XorA has many ;) Jan 03 13:02:26 greentux: thanks for asking Jan 03 13:02:32 ok, but You are @fosdem? Jan 03 13:02:44 perhaps somebody need a 3200 / 3300 for testing purpose... Jan 03 13:02:44 greentux: yes :) Jan 03 13:03:04 greentux: 3300? Jan 03 13:03:16 3100/3200 sorry... Jan 03 13:03:56 ah no secret source from greentux then :-) Jan 03 13:04:34 XorA: no, :) Jan 03 13:06:06 I heard about a successor of the olympus R1000 Touch PDA... Jan 03 13:06:49 * XorA only has a measly three Zs Jan 03 13:08:42 lrg: no, my desktop system :-/ Jan 03 13:08:58 RP: you have a desktop system? Jan 03 13:09:33 hrw|work: yes, easier to work on for long periods than a laptop Jan 03 13:09:53 RP: there was rumours that you develop on spitz only ;d Jan 03 13:11:20 hrw|work: nah, he is hardcore, collie thumb keyboard only Jan 03 13:11:26 hrw|work: :} Jan 03 13:11:28 ok, If somebody helps to have a zaurus for some months, we can org. it at fosdem Jan 03 13:11:52 hardes would be to develope on neo phone... Jan 03 13:12:06 greentux: but neo phone has 128M ram.. Jan 03 13:15:20 hrw|work: but You cant type with RAM ;) Jan 03 13:17:22 use it with voice controll on BT headset and speak the codez Jan 03 13:19:24 mickey is watching tv... Jan 03 13:19:44 XorA: with BT-less device you want to use BT headset? Jan 03 13:21:05 hrw|work: ah well, that blows that plan Jan 03 13:21:15 hrw|work: shoulda checked the specs Jan 03 13:21:31 ~seen zecke Jan 03 13:21:58 have to be a keyboard via SPI via micro-sd slot then Jan 03 13:22:05 zecke was last seen on IRC in channel #oe, 18h 57m 58s ago, saying: 'later guys'. Jan 03 13:23:18 XorA: micro-sd is under battery. :) Jan 03 13:24:10 stefan_schmidt: not a problem, I used to work for a company used to make cables for putting into SIM slots for simulations Jan 03 13:24:24 stefan_schmidt: its quite easy to route round a battery Jan 03 13:24:30 XorA: heh Jan 03 13:25:29 I'm using the linux-rp_2.6.19 kernel and it is having a core dump everytime I run "ipkg install" Jan 03 13:25:32 * stefan_schmidt still prefers BT keyboard. (Yes, I'm confident it will have BT) Jan 03 13:26:25 has anyone else had this issue? Jan 03 13:27:32 * XorA still needs to upgrade to 2.6.19, but last time I tried it I got crashes on boot when locales go OOM Jan 03 13:32:39 I'm getting an out_of_memory and a paging error in the dump. Jan 03 13:36:02 XorA, what version are you using? Jan 03 13:36:38 Gerrath: Im not at moment, I tried it when trying to find heisenbug Jan 03 13:39:22 The 2.6.19 kernel is crashing everytime I run "ipkg install", "ipkg update" and "ipkg remove" work fine. I will try 2.6.18 and see if it works better, I know 2.6.17 works fine.. although I have added all the RP patches, so maybe something there is not liking ipkg or is conflicting with one of my other patches :-/ Jan 03 13:44:45 I did an strace on ipkg and it looks like it is crashing during the download. Jan 03 14:16:32 Gerrath: may want to run some tests w/ wget or whatever ipkg uses to download files. Jan 03 14:17:14 ipkg update -verbose-wget and you will see Jan 03 14:18:11 Gerrath: is CONFIG_PREEMPT set in this version of the kernel? Jan 03 14:21:15 cbrake, no. Jan 03 14:23:24 actually I think the dump is happening before the download.. let me strace agaiun. Jan 03 14:33:35 niv_one_three: ping Jan 03 14:37:34 hmz Jan 03 14:38:47 cbrake, I'm removing all NON-critical patches for my system and re-compiling the kernel.. I will see if the issue is still there and if not then add the patches back a couple at a time. Jan 03 14:39:25 cbrake, once I identify the area causing it then I can litter it with printk's to find the source. Jan 03 14:39:46 why does makedevs not what it tells you it does? ie. it does not create directories Jan 03 14:45:24 and it ignores the uid/gid Jan 03 14:49:54 mr_mice : hi Jan 03 14:50:41 niv_one_three: hi Jan 03 14:51:16 niv_one_three: wich simpad do you have? Jan 03 14:51:23 pls jabber me: nivwiz at gmail.com Jan 03 14:51:29 SL4 Jan 03 14:51:38 u? Jan 03 14:51:50 help Jan 03 14:51:56 ~help Jan 03 14:52:00 niv_one_three: I don't have jabber but if you want we can do a private chat here Jan 03 14:52:30 ok Jan 03 14:53:32 niv_one_three: mom I have to read the IRC manual Jan 03 14:54:05 Sorry for the newb question, but does bitbake keep a central log file somewhere? Jan 03 14:54:46 hvontres|poodle: log of? Jan 03 14:54:48 wow, who made this madedevs thing? Jan 03 14:54:51 niv_one_three: do you konw how to start a private chat in irc? Jan 03 14:54:57 it's a wonder it works at all Jan 03 14:55:17 mr_nice: /query NICK Jan 03 14:55:23 mr_mice : to chat privatly usually /msg msg Jan 03 14:55:32 hrw|work: thx Jan 03 14:57:02 niv_one_three: do you see the message? Jan 03 14:57:18 I see yu at #oe Jan 03 14:57:29 the private msg also arrived Jan 03 14:57:35 the previos msg Jan 03 14:57:41 hrw|work: I was wondering if the screen output was also logged to a file by default. I started a build at home last night and I want to keep an eye on it from work Jan 03 14:59:37 hvontres|poodle: tmp/work/*/temp/log.do_* Jan 03 15:00:09 hrw: once the build is done , where do you see the .jffs2 image? Jan 03 15:00:17 hrw|work: thanks. I was wondering if there was another location :) Jan 03 15:00:24 ~/devel/org.embedded ? Jan 03 15:00:39 * hvontres|poodle just keeps an eye on top for now :) Jan 03 15:00:46 hvontres|poodle: use screen? Jan 03 15:01:02 detatch it and re-attach from work? Jan 03 15:01:39 niv_one_three: isn't there something like mr_nice like a chanel? Jan 03 15:01:41 a-atwood: hmm.. I guess I'll have to take a look at the screen man pages :) Jan 03 15:03:19 niv_one_three: tmp/deploy/images/ usually Jan 03 15:03:30 hrw: thanks Jan 03 15:03:51 nice: I do see a channel with you nick Jan 03 15:04:00 go on respond to my lines Jan 03 15:04:51 niv_one_three: I see a channel with your nick but no questions from you in it Jan 03 15:06:29 niv_one_three: did you saw my messages in that chanel? else we can talk in this chanell it it is not to ot Jan 03 15:07:18 niv_one_three: I hava a SL4 and a T-Sinus Jan 03 15:09:16 niv_one_three: you said that you have a problem with zeckes 2.6 mq200 driver. what exactly is your problem with it? Jan 03 15:24:55 03cbrake 07org.oe.dev * rb0ce4eb6... 10/ (1 MAINTAINERS): update MAINTAINERS file Jan 03 15:25:01 03cbrake 07org.oe.dev * rf53f1f02... 10/ (1 conf/machine/compulab-pxa270.conf): compulab-pxa270.conf: corrections in machine conf Jan 03 15:25:06 03cbrake 07org.oe.dev * r93bfcac1... 10/ (1 packages/linux/compulab-pxa270_2.6.16.bb): Jan 03 15:25:07 linux/compulab-pxa270: add md5sum check and note about binary NAND driver Jan 03 15:25:07 - md5sum check added because compulab does not version their downloads Jan 03 15:29:35 what are the requirements to become an official OE developer? Jan 03 15:29:52 Gerrath: submit good patches Jan 03 15:30:37 hrw|work, simple enough :-) Jan 03 15:31:23 Gerrath: you submit good patches, we give r/w Jan 03 15:31:32 simple Jan 03 15:57:52 Gerrath: You have the xscale_workaround patch? Jan 03 15:58:07 hi mallum :) Jan 03 15:59:20 XorA: You should try a kernel with the xscale patch applied - that could solve the OOM problem Jan 03 16:00:49 source mirror with all stuff in one directory HURTS Jan 03 16:00:53 hey RP :) Jan 03 16:01:04 hi Crofton mallum Jan 03 16:01:12 [hrw@misc1 ~]$ ls -1 www/sources/|wc 22830 22830 1241178 Jan 03 16:01:17 gm Jan 03 16:01:34 hey hrw|work Jan 03 16:01:51 RP, is that one of your patches? Jan 03 16:03:59 RP: I have one compiled up, just need to flash it and boot it Jan 03 16:07:20 how to get first letter from string in shell? Jan 03 16:07:35 name = 'umbaumba' and I want first 'u' Jan 03 16:09:02 RP is that the xscale_cache_workaround-r1.patch Jan 03 16:09:19 sed -e "s/(.).*/\1/" Jan 03 16:09:54 cut -c1 Jan 03 16:10:10 RP, if so I do not have it (yet), could that be the cause? Jan 03 16:10:41 thx gremlin[it] Jan 03 16:12:53 re Jan 03 16:20:03 cu guys Jan 03 16:26:02 Gerrath: It might fix the problem you're seeing. Its in mainline now. See the dicussion on linux-arm-kernel about it Jan 03 16:29:46 zecke: hi Jan 03 16:30:48 RP: I'm still thinking but forced to share my mind with the minix assignment... Jan 03 16:31:43 assignement : write a better OS than minix. *LOL* Jan 03 16:32:21 gremlin[it]: even better? Jan 03 16:33:53 is really good minix ? Jan 03 16:34:26 minix is easy to cope with Jan 03 16:34:32 no races, no threading :) Jan 03 16:34:46 zecke: good afternoon Jan 03 16:35:04 gremlin[it]: and last. Linux is obsolete, I use minix Jan 03 16:36:44 zecke: do you remeber if the 2.6 simpad_switches did work? Jan 03 16:36:52 zecke: when will we see minix_cvs.bb ? Jan 03 16:37:44 mr_nice: at one point it did. But nowadays pavel's mcp stuff is mainline and should be used Jan 03 16:38:46 zecke: yes but he has no switches or? Jan 03 16:44:17 zecke: do you know how to read out all necessary registers of a running linux box or how to find some docu or google keywords for it? Jan 03 16:46:47 zecke: I will compare the SIMpad with a functional mq200 state (after booting 2.4 and rebooting 2.6) and the same in a state where it won't work (after disconnect the power and boot 2.6 without booting ce or 2.4) Jan 03 16:47:01 mr_nice: arm linux, look at the ARM Linux Kernel ML FAQ (devmem2) Jan 03 16:47:17 'bitbake devmem2' Jan 03 16:47:19 zecke: thx Jan 03 16:50:07 mickey|tv: http://www.zdnet.com.au/news/software/soa/Long_term_Fedora_Linux_support_ending/0,130061733,339272842,00.htm Jan 03 16:52:30 gremlin[it]: when new note on th blog? Jan 03 16:54:58 univac, hi !!! ... as soon i'll got some result on USB ... Jan 03 16:55:14 Hi, o, nice! Jan 03 16:58:57 zecke really a shane fedora Legacy will close ... i update as a new FC exit my laptop and workstation ... but other small machines stay with old since big issue emerge (i have a couple FC4 small servers) ... Jan 03 17:00:05 zecke: hi and np :) Jan 03 17:12:19 gremlin[it]: common Free Software theme, people consume Jan 03 17:30:03 03koen 07org.oe.dev * r1c76e8be... 10/ (1 packages/fbgrab/fbgrab_1.0.bb): fbgrab: add patch to enable a network mode, see http://lists.openezx.org/pipermail/openezx-devel/2007-January/000941.html for details Jan 03 17:31:51 RP: do you know why the kernel seems to ignore the "-j 2" option for make and run single-threadded? Jan 03 17:32:33 hvontres|poodle: Known mainline kernel bug Jan 03 17:33:44 RP: OK..thanks. I am building a test image for poodle with 2.6.19+git atm. I should have some test results by friday :) Jan 03 17:34:43 03koen 07org.oe.dev * r90c738a1... 10/ (1 conf/machine/h1940.conf): h1940: tune for arm920t Jan 03 17:35:06 hrw|gone: there you go, no more arm packages for h1940 Jan 03 17:36:05 RP: can I do 'PACKAGE_ARCH=arm-oabi' and 'EXTRA_PACKAGE_ARCHS += "arm-oabi"' in a distro.conf? Jan 03 17:38:23 re Jan 03 17:38:24 koen: ok Jan 03 17:39:20 ipkg-make-index takes 26minutes Jan 03 17:39:20 ARHG Jan 03 17:39:21 i-m- drives me nuts Jan 03 17:39:35 ERROR: Cannot satisfy the following dependencies for task-base: kernel-image-2.6.15gum kernel-image-2.6.15gum kernel-image-2.6.15gum kernel-image-2.6.15gum kernel-image-2.6.17.9 kernel-image-2.6.1 Jan 03 17:39:39 5gum kernel-image-2.6.15gum kernel-image-2.6.18 kernel-image-2.6.17.9 Jan 03 17:40:20 * koen suspects linux-h1940 needs an eabi switch as well Jan 03 17:40:33 03koen 07org.oe.dev * r7e82eca3... 10/ (1 conf/machine/hx2000.conf): hx2000: clean up machine config a bit Jan 03 17:40:57 koen: thats for ks* machine Jan 03 17:50:58 ~oebug 1749 Jan 03 17:51:03 !oebug 1749 Jan 03 17:51:04 * * Bug 1749, Status: NEW, Created: 2007-01-03 10:22 Jan 03 17:51:05 * * openembedded(AT)hrw.one.pl: mc extfs contain STAGING related paths to perl Jan 03 17:51:06 * * http://bugs.openembedded.org/show_bug.cgi?id=1749 Jan 03 17:54:52 koen: I just differ recipes of bootstrap-image and angstrom-bootstrap-image. for my test both are ok. Jan 03 17:56:43 strace and gdb have had build problems for various archs Jan 03 17:56:56 so angstrom-*-image are better tests Jan 03 17:58:37 arm armv4t armv6 armv5te i586 i686 built here Jan 03 17:58:42 6.4/6.6 Jan 03 18:02:30 * mrz80 is away: Metaphysically I'm still here; physically, I'm elsewhere Jan 03 18:04:06 mrz80: metaphysically you are in queue to be added into ignore list Jan 03 18:05:20 hehe Jan 03 18:25:12 rp, xscale_cache_workaround-r1.patch did not fix my crashing. Jan 03 18:57:23 NCIS time! Jan 03 19:04:33 * hrw want faster ipkg-make-index Jan 03 19:05:46 hm. I wrote shell version but it was even slower Jan 03 19:06:26 cu Jan 03 19:20:19 * mrz80 is back (gone 01:17:50) Jan 03 19:24:05 ping do13 Jan 03 19:24:45 greentux: pong Jan 03 19:25:19 do13: evening... did you talk with your collegue? Jan 03 19:26:35 greentux: hi, yep no longer interested. another option :) Jan 03 19:27:29 do13: oh.. ok these times are good for job hoppers Jan 03 19:28:01 greentux: yep Jan 03 19:29:37 greentux: my collegue told me: this offer for the office in Essen would be more interesting Jan 03 19:30:27 do13: aahhh. Jan 03 19:30:50 do13: these "go west" people... Jan 03 19:33:40 greentux: you are from dresden? Jan 03 19:33:41 do13: is essen we have only student jobs open ;) Jan 03 19:33:51 do13: yes I am (radebeul, near DD) Jan 03 19:34:46 greentux: yep. He's looking for an job offer as hardware developer near Essen Jan 03 19:34:54 do13: do you live in/near Stuttgart ? Jan 03 19:35:07 hey Liam Jan 03 19:35:14 lrg: no Jan 03 19:35:32 do13: ah, fwiw we now have office in Stuttgart Jan 03 19:36:24 lrg: ahh ... Jan 03 19:36:51 do13: it's a 2 man office atm, but it will hopefully expand. Jan 03 19:37:17 lrg: sales? Jan 03 19:37:24 hehe Jan 03 19:37:33 do13: no, eng :) Jan 03 19:40:19 lrg: option for teleworking? :) Jan 03 19:41:40 do13: it's possible. Jan 03 19:41:51 lrg: btw. I have a new job in 3 weeks Jan 03 19:42:49 Why all the changes in oe.dev? There where 251 revisins since yesterday . Jan 03 19:43:03 goxboxlive: see mail, bogus merges Jan 03 19:43:21 zecke: ok Jan 03 19:43:28 do13: I'm curious... Jan 03 19:45:09 zecke :) Jan 03 19:47:42 zecke: telecoms protocol analysing (ATM, GSM/GPRS/UMTS and CDMA) Jan 03 19:48:17 mickey|tv: http://mail.gnome.org/archives/gtk-devel-list/2007-January/msg00018.html Jan 03 19:52:52 re Jan 03 19:53:32 hi Marcin Jan 03 19:58:46 68G devel/build/test/ Jan 03 19:58:46 real 232m37.373s Jan 03 19:58:46 user 0m6.853s Jan 03 19:58:46 sys 0m51.773s Jan 03 19:59:07 nearly 4h to 'just' check 'du -h' of my build... Jan 03 19:59:21 awesome Jan 03 19:59:32 ~lart slow serial-ata disc Jan 03 19:59:32 * ibot judo chops slow serial-ata disc Jan 03 20:08:06 re koen Jan 03 20:09:01 hey hrw Jan 03 20:09:14 hrw what dick is ... so i don't buy one :) Jan 03 20:09:21 s/dick/disk/ Jan 03 20:13:40 Vendor: ATA Model: Maxtor 6L250S0 Rev: BANC Jan 03 20:13:40 Type: Direct-Access ANSI SCSI revision: 05 Jan 03 20:13:40 SCSI device sda: 488397168 512-byte hdwr sectors (250059 MB) Jan 03 20:14:29 night all Jan 03 20:14:35 4h for checking... I wonder how long it will remove those files Jan 03 20:14:39 night lrg Jan 03 20:26:09 do13: interesting Jan 03 20:26:36 good nite Jan 03 20:42:27 for efika use distro angstrom-2007.1? Jan 03 20:42:35 will work Jan 03 20:42:41 according to koen Jan 03 20:42:47 RewriteRule ^(.)(.)(.*)$ $1/$1$2/$1$2$3 [L] Jan 03 20:42:54 gcc-3.4.4.tar.gz -> g/gc/gcc-3.4.4.tar.gz it should do or not? Jan 03 20:42:57 ideas? Jan 03 20:43:28 RP: can you look at my recent bitbake-dev mail? Jan 03 20:45:31 hrw: I'm ok on the licence change for 1.6, the setup.py and shbang changes. Please don't check that runqueue one in though Jan 03 20:46:17 RP: that one is to ignore Jan 03 20:46:58 hrw: the rest is fine. Just please make sure trunk sees the changes as well as 1.6 Jan 03 20:47:11 I've like to get a 1.6 release out soon Jan 03 20:47:46 ok. will do them tomorrow Jan 03 20:48:27 I backported enough from trunk to make MACHINE work from the environment too Jan 03 20:48:34 I saw that Jan 03 20:49:27 mc + 22K of files == long wait... Jan 03 20:50:12 :) Jan 03 20:50:35 * RP hugs mc Jan 03 20:51:36 ~lart clucene_cvs.bb for module=. Jan 03 20:51:37 * ibot calls clucene_cvs.bb on the phone ... the lights are on but nobody's home for module=. Jan 03 20:51:46 its HIDDEN in dl_dir Jan 03 20:53:38 note to self: Spell efika properly Jan 03 20:56:00 cu Jan 03 20:58:18 anyone know if there an existing cfengine-like package in OE? Jan 03 20:58:23 can't get cfengine to crosscompile properly :( Jan 03 21:38:51 koen: such a meeting sounds good, however the 23th would conflict with the Roy D'Espagne gathering Jan 03 21:39:04 drat Jan 03 21:39:04 koen: perhaps we could suggest the 26th? Jan 03 21:39:29 * koen wanted to jump up and down while shouting 'embedded' Jan 03 21:39:36 mickeyl: yeah Jan 03 21:40:05 * mickeyl wants to jump up and down while shouting "GtkBuilder and Widget Factory" Jan 03 21:40:09 bbiab Jan 03 21:47:49 RP: ping Jan 03 21:49:53 mickeyl: pong Jan 03 21:50:24 heya. I have a spare hour to hack on the xmlrpc server. can you recapitulate what waitpid functions we want in the server request loop? Jan 03 21:51:04 mickeyl: yes, I can spend some time on that now Jan 03 21:51:39 ok. i just need to know what i should do in the server loop and what API you want (i.e. addWaitPid(pid, callback) and then I'll do it Jan 03 21:52:17 mickeyl: Basically what the code at the end of runqueue does, just the waitpid there Jan 03 21:52:39 does one server need to wait for multiple PIDs ? Jan 03 21:53:17 mickeyl: waitpid will return any pid of any exitting process. We would then call it again to get the next one. Jan 03 21:53:48 mickeyl: We wait for any process in the current process to exit, then process that result Jan 03 21:53:55 current process group Jan 03 21:54:06 ok, so we wait nonblocking for 500ms ? Jan 03 21:54:23 and then select 500ms for incoming xmlrpc events Jan 03 21:54:24 ? Jan 03 21:54:36 * mickeyl inspects the runqueue code Jan 03 21:55:03 mickeyl: just call waitpid with no delay, then wait for xmlrpc for 500ms, then call waitpid, see if anything changed, no delay Jan 03 21:55:14 * likewise is hacking on lighttpd w/ fastcgi w/ xmlrpc and wonders what mickeyl and RP are developing? Jan 03 21:55:34 mickeyl: xmlrpc events are going to be much more time critical than process changes Jan 03 21:55:41 likewise: bitbake Jan 03 21:56:04 RP: ok, giving it an XMLRPC i/f? Jan 03 21:56:36 RP: k, let me cook up something Jan 03 21:57:53 likewise: something like that Jan 03 22:00:53 * mickey|bitbake digs out ElanMultipleSocketWatcher Jan 03 22:02:19 * RP fires up his nslu2 Jan 03 22:07:31 ~seen gerwinin Jan 03 22:07:57 gerwinin was last seen on IRC in channel #oe, 1d 2h 37m 18s ago, saying: 'Koen : didn't have to stay in the hospital'. Jan 03 22:08:03 * mickey|bitbake still has no turboslug Jan 03 22:08:07 *sigh* Jan 03 22:08:47 I did that as soon as I got it. I never got a serial lead working until about 5 minutes ago though :) Jan 03 22:08:58 heh, congrats Jan 03 22:08:58 hardware hacker Jan 03 22:09:11 i'm shy of even shortcutting a resistor Jan 03 22:09:26 First time I did the wiring correctly, I just picked up the wrong chip off the desk :-( Jan 03 22:09:36 eeks Jan 03 22:09:58 gah, you guys would start talking slug, look who you dragged in :-) Jan 03 22:10:10 heh, talking about the devil Jan 03 22:10:29 rwhitby: I finally have a serial console on a slug ;-) Jan 03 22:11:25 RP: sweet Jan 03 22:11:33 now you can help us with Angstrom :-) Jan 03 22:12:13 rwhitby: yes :) Jan 03 22:13:17 and maybe even the open source ixp npe driver bugs ... Jan 03 22:14:19 rwhitby: We'll see :) Jan 03 22:17:27 * koen proposes to clone the oe core team 10 times Jan 03 22:18:22 * mickey|bitbake supports the proposal Jan 03 22:18:29 what modules have to be loaded for usbfs to mean? Jan 03 22:18:31 * RP too Jan 03 22:18:37 usb-core or ohci* ? Jan 03 22:19:12 koen: It will probably mount with just usb-core Jan 03 22:19:34 It'll probably be empty until you load an hcd driver Jan 03 22:19:54 any objections againts 1) building usb-core into the kernel or b) autoloading usb-core ? Jan 03 22:20:23 koen: 1 will break the rootfs size on zaurii Jan 03 22:20:32 zimage size Jan 03 22:20:59 so b) Jan 03 22:21:04 no problem with autoloading on those machines with usb Jan 03 22:26:41 like this: http://rafb.net/p/UaQprI66.html ? Jan 03 22:29:24 mickey|bitbake: one of the irex dudes is coming to fosdem Jan 03 22:36:16 *lightbulb* Jan 03 22:36:31 with machine-via-env I can do builds for multiple machines at once Jan 03 22:37:08 can someone sponsor a 8 or 16 core machine + rackspace? Jan 03 22:37:18 koen: That patch is going to break things like c7x0 :-/ Jan 03 22:37:28 RP: how so? Jan 03 22:37:35 c7x0 doesn't have that module installed Jan 03 22:38:01 koen: I guess it isn't in the default rootfs. It usually is in mine though :-/ Jan 03 22:38:21 what ill effects does loading usb-core have? Jan 03 22:38:38 Its just pointless on c7x0 unless you have a usb CF card Jan 03 22:38:43 (I have it on my machines as well for the ratoc card) Jan 03 22:39:26 Why do we need this? So the mount of usbfs works? Jan 03 22:39:31 yes Jan 03 22:40:49 * RP wonders about setting autoload based on MACHINE_FEATURES Jan 03 22:41:01 I spotted the sa1100-rtc module inthere, so I thought it would be a good place Jan 03 22:41:11 * koen wonders about the same Jan 03 22:44:08 phew Jan 03 22:44:17 RP: you got mail Jan 03 22:44:28 bitbake hackin' session finished Jan 03 22:44:29 ;) Jan 03 22:45:08 mickeyl: thanks Jan 03 22:45:10 * RP reads Jan 03 22:45:21 np, it was about time. sorry [as always] for the delay Jan 03 22:45:23 :) Jan 03 22:47:51 * mickeyl ponders whether to buy a Laserpod[tm] Jan 03 22:47:53 http://www.laserpod.com/introduction.asp Jan 03 22:48:00 mickeyl: np :) Jan 03 22:49:06 mickeyl: This integrates the waitpid problem nicely but I'm still a little confused regarding data flow. Jan 03 22:49:09 mickeyl: hippie Jan 03 22:49:34 RP: in what way? Jan 03 22:49:36 koen: :) Jan 03 22:50:01 mickeyl: How do we handle callbacks from the server to the UI? Jan 03 22:50:13 mickeyl: get one of these: http://www.new-line.nl/webshop/view.asp?i=13913 Jan 03 22:50:32 mickeyl: Have you got the data flow email I sent handy (8th Decemeber)? Jan 03 22:51:15 RP: please send it to me again, i only have my response :/ Jan 03 22:51:53 sleep tight, good nite Jan 03 22:51:54 RP: I'm going to add some server bits into the ncurses to demo it for you, ok? Jan 03 22:51:59 (not today though) Jan 03 22:52:05 koen: do you know how to find documentation about usefull memory adresses for the devmem2? or other usefull docs or google keywords? Jan 03 22:53:07 RP: basically the GUI needs to open up a server as well Jan 03 22:53:15 integrated into the ncurses loop Jan 03 22:53:23 so we'll need a similar trick as integrating the waitpid Jan 03 22:53:25 thinking about it Jan 03 22:53:31 i may solve it a bit more generically Jan 03 22:53:46 to make it useful both for waitpid and ncurses Jan 03 22:54:00 * mickeyl will do that tomorrow Jan 03 22:54:04 bed time now Jan 03 22:54:05 g'night Jan 03 22:54:12 mickey|zzZZzz: one second Jan 03 22:54:15 * koen makes note to harass mickey|zzZZzz for screenshots Jan 03 22:54:20 ya Jan 03 22:54:22 mickey|zzZZzz: To quote the mail I just sent, " We don't want an xmlrpc server on both ends do we?" :) Jan 03 22:54:28 heh Jan 03 22:54:29 we do Jan 03 22:54:36 If we need a server on both ends, I understand :) Jan 03 22:54:41 good :) Jan 03 22:54:54 mickey|zzZZzz: great, thanks :) Jan 03 22:55:03 * koen follows mickey|zzZZzz example Jan 03 22:55:06 'night all Jan 03 22:55:07 np. pester me if I don't send you something tomorrow Jan 03 22:55:09 g'night koen Jan 03 22:55:18 'night mickey|zzZZzz and thanks again Jan 03 22:55:32 not at all :) Jan 03 22:55:58 03pfalcon 07org.oe.dev * r4b865f12... 10/ (1 conf/distro/angstrom-2007.1.conf): angstrom-2007.1.conf: Add feed arch for h3900. Jan 03 22:56:04 mickey|zzZZzz: I can probably make something work from here :) Jan 03 23:42:13 should we try selling OE stuff at FOSDEM? Jan 03 23:46:55 Crofton: What kind of stuff? Jan 03 23:49:01 stuff that people will buy Jan 03 23:49:08 t-shirts mouse pads etc Jan 03 23:50:14 Crofton: It could be good at raising funds but someone would need to organise it... Jan 03 23:50:22 yeah Jan 03 23:50:32 Unfortunatelu I am in the US Jan 03 23:50:48 dunno much about getting stuff like that made in US Jan 03 23:51:02 er EU :) Jan 03 23:51:49 Organising it will be a lot of work :-/ Jan 04 00:14:48 I guess we should see how much interest there is Jan 04 00:29:15 rwhitby: The size limit on a slug kernel is 1MB? Jan 04 00:29:30 RP: not any more Jan 04 00:29:47 (we use Apex as a second stage bootloader, and get past that limit) Jan 04 00:30:25 rwhitby: I'm seeing warnings about my kernel being too large when trying to reflash the device Jan 04 00:30:58 oh, the reflash script hasn't been updated yet to handle apex as a second stage bootloader Jan 04 00:31:07 if you upslug2 a complete 8MB image it will be fine. Jan 04 00:31:30 rwhitby: ah, ok, thanks Jan 04 00:31:44 (the reflash script just has to be taught how to insert the 16 byte sercomm header in the middle of the kernel image before it flashes it) Jan 04 00:32:10 the code is in the Debian flash-kernel script - we just need to copy it over and hack it a bit, Jan 04 00:32:11 hmmm...EABI fixes some alignment problems with gpsd's garmin support... Jan 04 00:38:38 rwhitby: Uploading an image just seems to loop with "download: GET_VERSION_INFO" appearing on the console repeatedly Jan 04 00:41:37 RP: uploading with upslug2 ? Jan 04 00:41:47 rwhitby: yes Jan 04 00:41:53 and it's in upgrade mode? Jan 04 00:42:37 rwhitby: LED flashing green/red, serial port reported Entry to download Jan 04 00:42:59 what's upslug2 saying? Jan 04 00:44:52 ~pastebin Jan 04 00:44:54 pastebin is, like, a place to paste your stuff without flooding the channel - try http://pastebin.ca, or http://channels.debian.net/paste Jan 04 00:45:24 rwhitby: http://pastebin.ca/304008 Jan 04 00:46:42 rwhitby: The ethernet light is flashing repeatedly, ready/status alternating read/green and the serial port printing "download: GET_VERSION_INFO" repeatedly... Jan 04 00:49:15 rwhitby: Wrong mac address :-(. 0 instead of an 8. Jan 04 00:49:23 * RP hangs head in shame Jan 04 00:50:01 if you only have one slug on the net, then you don't need to specify the mac address Jan 04 00:50:23 (upslug2 will choose the single slug in upgrade mode automatically) Jan 04 00:51:10 rwhitby: It was left in after I'd tried to upload the rootfs and kernel which I read needed that set Jan 04 00:52:18 hmm. it has a buzzer :) Jan 04 00:52:50 it's known as "the strangled beep" Jan 04 00:54:53 rwhitby: It surprised me slightly :) Jan 04 00:56:16 rwhitby: What's the status of angstrom on the device? Jan 04 01:11:17 RP: do you know much about how EABI changes things? Jan 04 01:11:25 JustinP: a little Jan 04 01:11:51 RP: I've figured out that gpsd on Angstrom (with EABI) works fine but on Angstrom compiled without EABI it has alignment problems and gives garbled data. Jan 04 01:12:05 RP: (OZ also gives garbled data) Jan 04 01:12:37 it's a binary protocol that is dependant on alignment and such, they use a union to read the data. With EABI I get no warnings. Without EABI I get warnings about alignment. Jan 04 01:12:58 RP: Angstrom rootfs is currently too big for NSLU2. We need to defeature it (starting by removing the zImage file from /boot). Jan 04 01:14:31 rwhitby: ok. I might be able to help with that :). Jan 04 01:14:32 bugger building angstrom-bootstrap-image failed on madwifi ... Jan 04 01:14:44 RP: that's what I'm counting on ;-) Jan 04 01:14:53 or efika Jan 04 01:15:37 JustinP: EABI did change lots of structure alignments. Are any enums involved? Jan 04 01:17:02 RP: let me check.... Jan 04 01:23:56 JustinP: EABI can optimise short enums (0-255) into a char instead of a long so any structures involving enums can change under eabi. I don't know if that applies or not but its caught me out before. Offhand, I'm not sure about alignments but they did change, you'd have to look that up Jan 04 01:24:01 * RP -> sleep Jan 04 01:24:03 'night a;; Jan 04 01:29:31 night RP Jan 04 01:32:23 RP: night. Any idea where I can look up these changes? **** ENDING LOGGING AT Thu Jan 04 03:00:02 2007